Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
A researcher at Alignment of Complex Systems showed that a 32B open-source model, Qwen2.5-Coder-32B, can subtly introspect when external concepts are injected into its activations, despite appearing unable to do so. By a…
A developer found that DSPy's compiled programs re-send up to 20 few-shot demonstrations on every inference call, adding significant token overhead that is invisible in the source code. The default optimizer attaches 4 b…
A developer's final part of a CUDA/ROCm optimization series details implementing Mixture of Experts (MoE) inference, including a custom top-k gating kernel, expert parallelism with All-to-All communication, and CPU offlo…
Moderna and Merck's experimental mRNA cancer vaccine intismeran autogene met its primary endpoint of recurrence-free survival and a key secondary endpoint in a Phase 3 trial of 1,137 patients with resected stage IIB-IV m…
A developer shared on Hacker News that they trained a 125M parameter transformer-based language model to autocomplete piano melodies entirely on-device, with no internet connection required. The model tokenizes MIDI data…
A new CUDA kernel correctness harness compiles kernels with nvcc, loads them into Python via ctypes, and checks outputs against NumPy baselines, with timing via CUDA events. All 23 tests pass on an NVIDIA T4 GPU, coverin…
A developer argues that AI applications are software and should be tested with the same rigor as traditional code, advocating for evaluation datasets, prompt testing, and context testing. The developer highlights that pr…
Lino Avila and the JX Planet X project have released JX N-Body Engine 0.1.0, a Python toolkit for high-precision Newtonian N-body simulations with built-in numerical validation. The software uses a sixth-order Yoshida in…
A paper from Amazon Science presented at AAAI 2026 found that agentic keyword search achieves 94.5% of RAG's faithfulness score without a vector database, suggesting retrieval quality increasingly comes from the agent's …
Researchers propose MeanFlow-Transfer (MF-T) and Continuous Adversarial MeanFlow (CAMF) to adapt pretrained diffusion or flow models to new domains with limited data while accelerating sampling. MF-T maps heterogeneous s…
Researchers propose MS-WDRO, a multi-source Wasserstein distributionally robust graph learning framework that fuses heterogeneous source-domain data via a weighted Wasserstein barycenter and an ambiguity ball, yielding a…
A new arXiv paper (2608.20181v1) proposes a standardized framework for evaluating machine-learning-based power system protection, defining seven required study dimensions. In a case study on the PROTECT-90 benchmark with…
FleetSieve, a new profiling method for LLM serving fleets, reaches the oracle aggregate decision using 22,200 GPU-seconds, 6.9% less than uniform random profiling on a fixed H100 grid for a 31B-parameter model. Across 20…
A new arXiv paper (2608.19203v1) introduces Asymmetric Attention Heads (AAH), a framework that assigns different context window lengths to different attention heads in transformers, rather than giving every head the same…
Researchers from an unnamed institution report that a contrastive decoder can extract open-vocabulary word-level information from non-invasive EEG during silent reading, achieving reliable above-chance word-grouped top-1…
A new study from arXiv (2608.19218v1) proposes a time-series retrieval framework that grounds multimodal large language models (MLLMs) in remaining useful life (RUL) estimation, improving prediction accuracy on the C-MAP…
Researchers propose Answer-Level Trust Selection (ATS), a post-hoc framework that accepts or rejects individual vision-language model (VLM) predictions for physical quantity estimation without fine-tuning or internal log…
A new arXiv study (2608.19558v1) finds that confidence rankings for financial named entity recognition (NER) change under domain shift, with whole-output probability being the strongest in-domain error detector but deter…
A new arXiv preprint (2608.19497v1) characterizes the learning dynamics of a compact hybrid quantum forecasting model against a classical baseline, finding that both achieve similar held-out performance despite distinct …
Researchers proposed HiRA-CAM, an improved gradient-based visual explanation method for convolutional neural networks (CNNs), and showed it outperforms LayerCAM and Grad-CAM on creating saliency maps for object classific…